100% RE by 2050 and the Effects of Lifetime and Recycling

Comparison case using the functions in PV ICE to compare how PV module lifetime and PV module recycling effect the energy transition to 100% clean energy by 2050.

Folder 15 vs 50 year Module

Deployment Schedule File Preparation

NOTE: this section of code should only need to be run once to populate data, and again anytime the ReeDS file is updated.

First, we load the Module Baseline. This file will be used later to populate all the columns other than 'new_InstalledCapacity[MW]'. Deployment schedule for this analysis will be supplied by the REEDS model. This analysis will use PV ICE developed material and module baselines.

Drop 1995 through 2009 because chosen deployment schedule begins in 2010. Technically this neglects ~1.5 GW of installs from 1995 through 2009.

Now we load the deployment schedule. This is derived from ReEDS simulation output, and in this case we're using one of the deployment projections as used in Solar Futures (includes PCA regions, States).

NOTE: The ReEDS deployment file is stored outside of the PV ICE folder and therefore not publicly available on github. Please contact the authors for this file.

First create a copy which groups the data by PCA region

For each Scenario and for each PCA, combine with baseline and save as input file. This will be in a folder PCAs under the simulation folder in TEMP

For each Scenario and each State, combine with baseline file and save as input file. This will be in a folder States under the simulation folder in TEMP

Finally, make an overall US baseline which ignores PCA regions and states. This is useful for speeding the simulation and is the deployment schedule used in this analysis.

Analysis

The Solar Futures report has a variety of deployment scenarios. In this case, we are only concerned with the highest capacity and deployment rate, Decarbonization + Electrification (Decarb+E).

step: Collect all the scenario names and downselect to the scenario(s) of interest.

Set up the PV ICE simulation with scenario and materials

Run the simulation

Lifetime and Recycling Scenario Creation

The range of potential future technology directions for PV will be explored in terms of module lifetime and EoL recycling rates. Currently technology is ~32 year module with a 6% EoL recycling rate (15% collection, 40% of collected modules sent to recycling). Lifetimes could improve, with 50 years targeted by DOE SETO, and/or recycling rates could improve, as exemplified by CdTe management from First Solar or perovskite technology. This analysis will explore on a mass flow basis, which of these two circular economy levers (lifetime or recycling) is the most important research priority for achieving the energy transition while minimizing waste and material extraction.

We will explore from a 15 year module lifetime to a 50 year module lifetime, and from 0% recycled to 100% recycled.

Create lifetime and recycling ranges

Automatically generate T50 and T90 values for each lifetime, by creating a linear regression for mod_reliability_t50 & mod_reliability_t90 vs. mod_lifetime to estimate t50 and t90 values to input for various lifetimes.

Create all Scenarios

Now with the lifetime and recycling ranges defined, we will create a PV ICE scenario for each combination

Notes/Assumptions:

This method simplifies the coding, but can be representative of more complex systems (i.e. 50% collection rate with 80% recycling yield = 40% overall recycling rate.)

Now we run the scenarios. This run of the simulation uses identical installations (i.e. no replacements)

Use the PV ICE "aggregate results" function to print out a table of Virgin Material Demands, Lifecycle Wastes (MFG, EoL, both), new installed capacity and effective cumulative capacity, both annually and cumulatively.

Read the aggregated results back into the journal from csvs (run time on simulations can be long)

Figure 2

Here we select the data for figure 2. This is done by selecting a single column example of each lifetime, adding the PV ICE baseline, selecting for the effective capacity column, and printing to csv for graphing.

Recycling rate doesn't have an effect on effective capacity, only on the virgin material demand and lifecycle wastes, therefore we can select a single lifetime example and it doesn't matter which recycling rate it is.

Heat Map - Identical Installs

This Heatmap is not used in the publication. There are two graphs, virgin material demand and life cycle wastes. The X-axis is module lifetime and the y-axis is Recycling Rate. The colors are relative to the virgin material demand and life cycle wastes of the PV ICE Baseline. These graphs are for identical installations, meaning that shorter lifetime scenarios do not meet energy transition capacity targets

Figure 7 Pie Charts

Printing out the data for pie chart of cumulative Virgin Material Demands and Lifecycle Wastes in 2050, PV ICE scenario

Installation with Replacements - Compensation Calculation

This section of the analysis runs the capacity compensation simulation. For each scenario, each year the deployment "new installed capacity" is adjusted (up for short-lived modules or down for long-lived modules) by the difference in effective capacity between the scenario and the PV ICE Baseline effective capacity.

NOTE: this mass flow calculation takes a LONG time to run, recommend leaving it overnight. A csv of the yearly and cumulative aggregated results is saved as csv and read back in to speed analysis and graphing.

Read the csvs back in for plotting (installation compensation calc runs a LONG time).

Figure 3

Bar chart of additional installations

Figure 4 Heat Maps with Replacements

This creates the data frames and heat maps of virgin material demand and lifecycle wastes for all 336 scenarios, including replacement modules. Colors are relative to the PV ICE baseline.

First, pull out the material demand and wastes for the PV ICE Baseline

Figure 5 - Time Shifted Demand and Wastes

This code exports the data for Figure 5. The 0% recycling scenario is selected to represent the full material demand (since all other recycling scenarios offset the virgin material demand with recycled content). This allows comparison of overall material demands, and a visualization of how much of the lifecycle wastes could offset demand and when.

Table 1 - Select results

Print out table 1 select results for virgin material demand and lifecycle wastes.

Sanity Check: BOM decrease and Efficiency increase

These results were presented at PVRW 2022.

BOM modification

First method of approximating a "thin film" BOM as just glass, backsheet, an Aluminum Frame. This mass is similar to CdTe modules. A second method of reducing mass to thin film design is presented below.

Now calculate installation compensation for the 15 year module.

Search for the cumulative value that is less/more than the PV ICE baseline with all materials.

Here, we compare the PV ICE c-Si virgin material demand to a thin film of 15 year life. These results indicate that lowering the BOM will lower the required closed-loop recycling rate to reduce virgin material demands from 95% to 75%.

This compares the thin film BOM waste to the PV ICE c-Si baseline waste. The recyling requirement for lowering waste is still quite high. Given that most of the waste is attributable to pre-2020 low open-loop recycleable modules, little can be done about the pre 2050 waste.

Efficiency Modifcation

Now that we have confirmed that decreasing the mass per module area will lower the required closed-loop recycling rate, lets check that increasing module efficiency will have the same effect. Currently, PV ICE baseline is 20% efficiency in 2020 and 25% efficient in 2050. Oberbeck et al 2020 expect 30% efficient tandem devices (perovskite + silicon). Therefore, we will use this as an approximation of an efficiency increase, and will apply it to the 15 year module.

First we check the change in deployed capacity, since modifying module efficiency will primarily effect the deployment (and as a result effect virgin mateiral demand). Installs are only dependent on life, not recycling, therefore we can select any of the recycling rates.

Because we deploy using MW, the difference from efficiency improvement will not appear in MW deployed but in the area of those MW deployed. Therefore, we will compare the area deployed as a proxy for # of modules and compare the tandem 30% efficiency against the c-Si 25% efficiency 15 year modules, and PV ICE.

Area comparison

If we approximate a module as 2 m^2 (current average, though CdTe series 6 modules are 2.47 m^2), then we can use the area deployed to calculate an approximate number of modules.

This graph shows the cumulative deployed area over time for the PV ICE baseline, the "15-year Tandem" device, which is the same BOM but higher module efficiency (30%), and the c-Si 15 year module with the same efficiency as PV ICE (25%). We see that the higher efficiency lowers the required deployment area. Interestingly, around 2038, PV ICE and the 15-year Tandem device cross, because the replacement requirement for the 15-year Tandem is higher than PV ICE 35 year module. Cumulatively, the Tandem device still requires more area deployment. Next let's look at what level of closed-loop recycling will drop the virgin material requirements.

These results indicate that due to reduced area deployed, the necessary closed-loop recycling is lowered to 65%. This indicates that virgin material demand is more sensitive to module efficiency than BOM (which lowered the closed-loop recycling need to 75%).

Like the BOM change, lifecycle wastes are not as sensitive to module efficiency changes, since much waste is due to pre-2020 modules.

Breakthrough Technology: Thin Film + High Efficiency

Finally, what if there is breakthrough technology which is 30% efficient and a thin film technology (i.e. low BOM).

Now run capacity compensation. This is where the difference really lies, in how many fewer modules can be deployed due to increased module efficiency.

The reduction in area deployed should be identical to the module efficiency improvement analysis above. Therefore, we will just look at the combined effect on virigin material demand and the necessary closed-loop recycling rate.

The combined lower BOM and higher Efficiency reduce the required circularity to 50% closed-loop.

ALTERNATE BOM MODIFICATION

In this BOM modification, we will try to more accurately represent a thin film technology. Parameters will be identical glass baseline, Aluminum Frame, Backsheet, 10% of the c-Si, 50% of the encapsulant, and neglect Ag and Cu. The glass and backsheet baselines account for increasing shares of glass-glass packaging. Silicon absorber will be reduced to a thin film thickness as a proxy, and the encapsulant will be cut in half, since typically, thin film uses 1 sheet of encapsulant instead of the 2 sheets c-Si uses. This will include manufacturing inefficiencies of these materials. Current wafer thickness is ~165 micron, and CdTe thin films are 10s of microns; therefore we will use 10% of c-Si mass per area.

Now we analyze the results

These results indicate that 80% closed-loop recycling is required for this thin film BOM to extract fewer materials. This is an overall lowering of the BOM by:

This is ~12kg/m2. For reference, a CdTe series 6 module is ~14kg/m2